A one-person practice needs seven things from practice management software: client records with deadlines and recurring jobs, a portal for document requests, engagement letters with e-signature, time and billing, client messaging, a few simple reports and a clean data export. AI summaries and draft replies are useful extras, not reasons to buy; check them last.
Most of what gets sold to solo accountants is built for teams of ten, so you can end up paying for permissions, capacity planning and multi-team dashboards you will never open. The sensible approach is to write down your own client mix, test each tool against it for a week, and treat anything labelled beta or "coming soon" as a bonus rather than a promise.
What changes at one, four and ten people
Feature lists look alike until you ask who needs what. A solo practitioner never assigns work to a colleague, so task routing barely matters. They do need everything that replaces a receptionist and an administrator: chasers, reminders, intake forms and a tidy audit trail. As headcount grows, the weighting flips.
| Firm size | What matters most | What you can ignore for now |
|---|---|---|
| One person | Deadline tracking, document requests, e-signature, billing, data export | Roles and permissions, capacity planning, multi-team reporting |
| Two to four people | Task assignment, shared inbox, review steps, time budgets per job | Firm-wide analytics, data warehousing |
| Five to ten people | Workflow templates with roles, work-in-progress reporting, client groups, integrations with the ledger | Enterprise single sign-on, custom data pipelines |
The practical consequence: a solo practice should shortlist on the first row and then check the price of moving up when the first employee arrives. Several vendors gate useful automation behind a middle or upper tier, so "cheap to start" and "cheap to grow" are separate questions.
The seven core checks, each with a test you can run in a trial
Each check below says why it matters, then gives a test to run during the free trial. If a vendor will not give you a trial, or limits it to a guided demo, that is itself a data point.
1. Clients, deadlines and recurring jobs
You need one record per client holding the contacts, the services you provide and the dates that matter, plus job templates that create the same checklist every year or month. The test: import ten real clients (names changed if you prefer), set up one yearly job and one monthly job, and ask the software to show you everything due in the next 30 days. If building the monthly job takes more than 20 minutes, imagine doing it for 140 clients.
Also check whether jobs generate themselves for next year once this year's closes. Manual re-creation is the quiet time sink that sole practitioners notice in month eleven.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy now2. Document requests and the client portal
This is the check that earns the software its fee. You want a checklist of requested documents per job, clients who can upload from a phone, and reminders that go out without you. The test: send a request to a friend or family member posing as a client, have them upload one wrong file and one right one, and see what you receive. Can you rename and file documents in one click? Do you get an alert only when something arrives, or a flood?
Check which plan includes automatic reminders. On Karbon's published plans, "automatic client reminders" are listed under Business, not Team, which matters when the entry price is what attracted you. If you also want a systematic approach to the chasing itself, the tutorial on chasing missing client records before deadlines covers the wording and timing.
3. Engagement letters and e-signatures
Look for proposals or engagement letters that the client can accept and sign inside the portal, with the signed copy filed automatically against the client. The test: produce a letter from a template, merge in a client name and fee, send it, sign it from a phone, and confirm the signed PDF lands in the right folder with a timestamp. If you already draft letters with an assistant, AI engagement letters for accountants shows how to keep the scope wording consistent.
4. Time, billing and payments
Even if you charge fixed fees, you want time records so you can see which clients are unprofitable. Check for recurring invoices, a payment link on each invoice and a way to bill several entities in one invoice if you have family groups. The test: raise a recurring monthly invoice for a bookkeeping client, pay it by card, and see how the payment fee appears. Card fees differ by vendor and are rarely on the headline price, so ask for the rate in writing.
5. Messages and email
You should be able to see a client's emails and portal messages in one place. Some tools sync your mailbox and file emails against clients; others offer a secure message thread inside the portal. The test: forward three real client emails into the system and check whether attachments are saved against the right client and whether replies stay in the thread.
6. Reports you will actually open
A solo practitioner needs about three views: what is due this week, what is waiting on the client, and what is finished but unbilled. Everything else is decoration. The test: ask the tool for each of those three views without building a custom report. If one needs a custom build, count the steps.
7. Getting your data out again
Practice software holds years of client history, so leaving must be possible. Ask what you can export (client list, notes, time entries, invoices, documents with folder structure) and in what format. Karbon's comparison table, for example, shows a "standard" data export on Team and Business and a custom one only on Enterprise, with a six-month reporting period below Enterprise, so ask what "standard" contains before you rely on it. The test: during the trial, export what you have loaded and open it in a spreadsheet. Missing notes or flattened folders at that stage are cheap to discover.
Three vendors read from their own pages in November 2026
These figures come from each vendor's own pricing page, fetched in November 2026 and quoted in USD. Plans change, so confirm on the day. I name these three because all three publish prices and turn up on small-firm shortlists, not because they are the only choices; other products exist and the same checks apply to them.
| Vendor and plan | List price | What the page says about fit |
|---|---|---|
| Karbon Team | $59 per user per month, paid annually ($79 monthly, but monthly billing needs a team of four or more) | Described as the essential tools for small firms; capped at 1,000 contacts and 3 teams; integrations limited to email and general ledger |
| Karbon Business | $89 per user per month, paid annually ($99 monthly, again for teams of four or more) | Adds automatic client reminders, task automation, the full integration ecosystem, API access and unlimited contacts |
| TaxDome Essentials | $800 a year per seat on a one-year commitment ($750 on two years, $700 on three) | Sold for solo users only; includes CRM, portal and app, e-signatures, proposals, payments and ready-made automations |
| TaxDome Pro | $1,000 a year per seat (one-year term) | Adds ledger integrations, AI-powered reporting, roles and a 30-day activity feed |
| Canopy Standard and Plus | Standard $74 and Plus $109 per user per month, billed annually ($93 and $136 on monthly billing); a Premium tier at $149 is marked coming soon | Every tier includes CRM, portal, documents and e-signature, recurring task workflows and billing; AI document renaming starts on Plus; tax workflow automation is a credit add-on from $34 per client |
One more detail on billing. TaxDome's page says all plans are billed upfront, so a solo practitioner pays $800 in one go. Karbon's $59 rate is also paid annually, and its monthly option starts at four users, so a one-person firm should expect an annual commitment from either. Canopy lists monthly prices too; ask whether a single user can take them before you count on that flexibility. The yearly totals sit within a few hundred dollars of each other, so cash flow and fit may decide it rather than headline price.
A quick sum on what the fee has to earn back
Suppose you value your time at $60 an hour (an illustrative figure; use your own). Karbon Team on annual billing costs 12 x $59 = $708 a year, so it has to save about 11.8 hours a year, or roughly an hour a month. TaxDome Essentials at $800 needs about 13.3 hours a year, or 67 minutes a month. Canopy Standard at 12 x $74 = $888 needs 14.8 hours, about 74 minutes a month. Karbon Business at 12 x $89 = $1,068 needs 17.8 hours, which is about 90 minutes a month. If your document chasing alone takes more than that, any of the four clears the bar on paper. The sum fails when the software saves nothing because you never switch the automations on, which is the failure mode I would watch for.
Where AI sits: built in, add-on or still on a roadmap
Vendor pages describe AI in very different states of readiness, and the distinction matters when you are paying today. The table sorts what I could read into "usable now" and "not yet", using each vendor's own labels.
| Vendor | Listed as available | Listed as beta, early access or coming |
|---|---|---|
| Karbon | Email thread summaries, AI-drafted replies and drafts from tasks, client briefs, work-item snapshots, suggested assignees, flagging missed time entries | Kai AI coworker and AI agents (early access), meeting notetaker (coming soon) |
| TaxDome | Automatic tagging and renaming of uploaded documents on every plan; AI-powered reporting from Pro | Atlas operations agent, open connectors for your own AI tools and a public API (all still listed as coming soon in November 2026, with a beta waitlist for Atlas) |
| Canopy | AI document renaming (Plus and above); smart tax prep (marked new); AI-assisted tax workflow add-ons | Coworker agent (beta), smart e-signature fields (beta), AI meeting notetaker (beta) |
Two other points from the pages. Karbon says its AI is powered by Azure OpenAI Service, that data stays within Karbon and is not used to train AI models, and that its AI features currently cost nothing extra on any paid plan. TaxDome says data processed by its AI is not sold or used to train public or third-party models. Both are vendor statements, so get them in writing for your own file, as the tutorial on using ChatGPT with client data explains for general assistants.
Test an AI summary on a real thread
The simplest check on any built-in summary feature is to feed it a thread where you already know the answer. Take an anonymised email chain with a client and use a prompt like this, either inside the practice tool or in a general assistant for comparison:
Summarise this email thread for the client file in five bullet points.
Include: what the client asked for, what I replied, any date or figure
mentioned, anything still outstanding, and who owes the next action.
If a date or figure is missing from the thread, say "not stated" and
do not guess.An illustrative output might read: "Client asked whether the rental income schedule can wait until March. Replied that the deadline is fixed. Client mentioned a $4,200 repair invoice, date not stated. Outstanding: repair invoice copy. Next action: client." Now check it against the thread. Common faults are a softened deadline ("can probably wait"), an invented date for the repair, or a missed owner of the next action. If the tool gets all five points right on three different threads, the feature is saving you real filing time. If it needs correction every time, it is a drafting aid, not a record-keeper.
A worked scoring: 140 clients, one accountant
Here is an illustrative practice. One accountant looks after 140 clients: 85 individuals with personal returns, 30 small companies needing year-end accounts, and 25 monthly bookkeeping and payroll clients. Document chasing takes about 14 hours a week in the busy months and she wants it cut. She scores two unnamed tools out of 5 on weighted criteria, with AI deliberately set at 5 per cent.
| Criterion | Weight | Tool A score | Tool B score | A weighted | B weighted |
|---|---|---|---|---|---|
| Deadlines and recurring jobs | 20 | 5 | 3 | 100 | 60 |
| Portal and document requests | 20 | 4 | 5 | 80 | 100 |
| Engagement letters and e-signature | 10 | 4 | 4 | 40 | 40 |
| Billing and payments | 15 | 3 | 5 | 45 | 75 |
| Messages and email | 10 | 4 | 3 | 40 | 30 |
| Reports | 5 | 3 | 3 | 15 | 15 |
| Data export | 10 | 2 | 4 | 20 | 40 |
| AI features | 5 | 4 | 2 | 20 | 10 |
| Price fit | 5 | 3 | 4 | 15 | 20 |
| Total (out of 500) | 100 | 375 | 390 |
Tool B wins by 15 points, and the reasons are billing and data export, not AI. Tool A's better AI features were worth only 10 points of difference. That is how it should look: if a flashy assistant can flip the result, the weighting has gone wrong. She also notes that Tool A's weak export score (2) is a switching cost she would pay later, so she asks Tool A for a written description of what its export contains before ruling it out.
Before and after: the Friday chase
Before, she kept a spreadsheet of clients with missing records and sent a round of emails every Friday, about 45 minutes in a normal week and over three hours in January. Replies landed in her inbox with attachments named "scan0012.pdf". After moving to a portal with a document checklist, each job asks for specific items, reminds the client at set intervals and files uploads against the job. The Friday session becomes a five-minute scan of what is still outstanding. The saving only appears if reminders are switched on for each job template, so the first week's work is setting up three templates (personal return, year-end accounts, monthly bookkeeping), not moving clients.
A trial week that exposes the gaps
Give yourself five working days and treat the trial like a mini project. Each day has one job, and by Friday you will have scored every criterion from real use.
| Day | Task | What you are looking for |
|---|---|---|
| Monday | Import ten clients and build one yearly and one monthly job template | Time taken, fields that will not map, templates that regenerate next year |
| Tuesday | Send a document request to a friendly tester and have them upload from a phone | Client experience, file naming, whether reminders fire |
| Wednesday | Issue an engagement letter, sign it, raise an invoice and pay it | Signed copy filing, payment fee, receipt wording |
| Thursday | Run the AI features on three real, anonymised threads or documents | Accuracy, invented details, whether you can turn it off |
| Friday | Export everything, then ask support one awkward question | Export completeness and the quality of the reply |
The awkward question on Friday is a deliberate test. Ask something specific, such as what happens to data after cancellation or how long you would have to download it. A vague answer from a sales rep is information you will want before you hold client files in the system for three years.
The reminder feature that sat one tier up
An illustrative case: a sole practitioner buys the cheapest plan on annual billing because the entry price looks right, spends two weeks loading 120 clients, and only in the first busy month discovers that the reminders she wanted run on the next tier up. She now has two choices: pay the difference for the rest of the year or chase by hand. A single question before buying ("which plan sends reminders to clients automatically, and can I see it working on a trial of that plan?") would have saved the loading time. The same applies to seat rules: ask whether short-term or seasonal seats exist on your tier before you hire anyone for busy season. TaxDome, for example, lists monthly seats for short-term staff on Pro and seasonal seats on Business, not on Essentials.
If your wider worry is whether the whole AI stack is worth buying for a practice your size, AI for a one-person bookkeeping business gives a cost-led view, and the sibling tutorial on practice software automation compared looks at the automation differences between three vendors in more depth.
Questions to send a vendor before you pay
Put these in an email so you have the answers in writing. Fill in the bracketed parts for your practice; a filled-in version might look like this.
Hello [first name],
I run a one-person practice with about 140 clients (85 individuals,
30 small companies, 25 monthly bookkeeping). Before I buy, please
confirm in writing:
1. Which plan includes automatic client reminders and recurring jobs
that regenerate each year?
2. What can I export if I leave (clients, notes, time, invoices,
documents with folders), and in which format?
3. Are client documents or messages used to train any AI model? Which
provider processes them, and can I switch the AI features off?
4. What happens to my data if I cancel, and for how long can I
download it?
5. What is the card payment fee, and is it on the invoice or deducted?
6. What does moving to the next plan cost when I hire one person?
Thanks,
[your name]Keep the replies in the same folder as your supplier contracts. If the vendor answers three of the six vaguely, move to the next tool on the list. For help working through the supplier side of this decision with someone, my AI implementation consultation covers it; start at the contact page.
Scoring the AI extras without being swayed by them
Once the core checks are done, give each AI feature a simple three-way label: saves time now, saves time later, or marketing. A feature saves time now when you tested it on your own material and it needed light edits. It saves time later when it is described as early access, beta or coming, and you would revisit it in six months. It is marketing when no one can show it working on a document like yours.
For a practice your size, the features most likely to land in the first group are document tagging and renaming, email-thread summaries and draft replies to routine queries. These are small tasks, but they sit in the middle of your day. Features that act on your behalf, such as agents that set up new clients or send reminders on their own, deserve a different test: how do you see what they did, and can you approve before anything leaves? Karbon, TaxDome and Canopy all describe approval steps or action logs for their agent features, which is the right instinct; verify it in the trial rather than assuming it.
When the shortlist is down to one, write the three reasons you chose it on a page and file it. In a year, when someone offers a new AI feature, you will be able to ask whether it changes any of those three reasons. Usually it will not.
Questions solo accountants ask before paying
Do I need practice management software with fewer than 50 clients?
Probably not yet. Under about 50 clients, a shared calendar, a folder structure and a spreadsheet of deadlines can hold up. The tipping point is usually document chasing: when you spend more than a couple of hours a week asking for the same records, a portal with automatic reminders starts to pay for itself.
Is a plan labelled for solo users worth it?
Often yes, if it covers your must-haves. TaxDome's Essentials plan is sold for solo users only, which is cheaper per seat than its higher tiers. The risk is outgrowing it: check what it costs to move up when you hire, and whether seasonal seats are allowed on your tier.
Should AI features decide which tool I buy?
No. Several AI features on vendor pages are labelled beta, early access or coming soon, and the same tasks can be done with a general assistant. Score AI at about five per cent of your decision and spend your trial week on deadlines, portal, billing and data export.
What should I ask about client data and AI features?
Ask in writing whether client documents or messages are used to train any model, which AI provider processes them, whether you can switch the AI features off, and where the data sits. Keep the reply with your professional-indemnity paperwork, and check your professional body's rules on outsourcing client data.
Further reads
- How Much Does AI Cost a Small Accounting Firm? — Adds the AI line items a practice budget should carry.
- Best AI Tools for Small Accounting Firms, Sorted by Task (2026) — Tools grouped by task if you want add-ons, not a suite.
- AI Powered Bookkeeping Platform: How a Practice Chooses One — Same decision, but for the bookkeeping engine underneath.
- AI Receptionist for Accounting Firms: Tax-Season Calls and Cost — Covers the phone side once your portal is running.
- How to Read AI Software Pricing: Seats, Credits, and Usage Fees — Explains seats, credits and per-client fees before you sign.
- Hire an AI Consultant for an Accounting Firm: Scope, Cost, Red Flags — What to scope if you outsource the setup.
- Ten AI Use Cases in a Small Accounting Practice, Ranked by Set-Up Effort — Ten AI jobs for a small accounting practice, ranked by how much set-up and risk each carries, with an example and a first step for every one.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Karbon pricing and AI pages, TaxDome pricing and TaxDome AI pages, Canopy pricing page, all read in November 2026. Prices are USD list prices and change often.